#!/usr/bin/env bash
# Reproducible WebAssembly build of liblouis for uebtranslator.com.
#
# Inputs (all pinned):  liblouis release tarball (SHA-256 in liblouis-*.tar.gz.sha256),
#                       Emscripten SDK version, emcc-flags.txt, tables.list.
# Outputs (engine/out): liblouis.mjs, liblouis.<sha8>.wasm, tables/*, manifest.json.
#
# The library and tables are built and shipped UNMODIFIED. Nothing from liblouis-js (GPL-3.0)
# is used: no --pre-js/--post-js, our glue lives in src/lib/engine/ under MIT.
# Run only when the pinned version changes:  npm run build-engine
set -euo pipefail

LIBLOUIS_VERSION="3.39.0"
EMSDK_VERSION="6.0.10"
PYTHON_RELEASE="20260924"
PYTHON_VERSION="3.12.14"

HERE="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
SRC_DIR="$HERE/src"
# OUT_DIR and BUILD_DIR may be overridden: scripts/verify-engine-build.mjs unpacks, compiles and
# links in a temporary directory and compares hashes with the committed engine/out. The defaults
# are the committed output and engine/src (where the verified tarball is cached).
OUT_DIR="${OUT_DIR:-$HERE/out}"
BUILD_DIR="${BUILD_DIR:-$SRC_DIR}"
TARBALL="liblouis-${LIBLOUIS_VERSION}.tar.gz"
TARBALL_URL="https://github.com/liblouis/liblouis/releases/download/v${LIBLOUIS_VERSION}/${TARBALL}"
LIB_DIR="$BUILD_DIR/liblouis-${LIBLOUIS_VERSION}"

log() { printf '\033[1m[engine]\033[0m %s\n' "$*"; }
die() { printf '\033[31m[engine] %s\033[0m\n' "$*" >&2; exit 1; }
sha256() { shasum -a 256 "$1" | awk '{print $1}'; }

# ---------------------------------------------------------------------------
# 1. Source tarball, verified against the pinned hash
# ---------------------------------------------------------------------------
mkdir -p "$SRC_DIR"
if [[ ! -f "$SRC_DIR/$TARBALL" ]]; then
  log "downloading $TARBALL"
  curl -fsSL -o "$SRC_DIR/$TARBALL" "$TARBALL_URL"
fi
(cd "$SRC_DIR" && shasum -a 256 -c "$HERE/${TARBALL}.sha256") || die "tarball hash mismatch — refusing to build"
TARBALL_SHA="$(sha256 "$SRC_DIR/$TARBALL")"

rm -rf "$LIB_DIR"
mkdir -p "$BUILD_DIR"
tar xzf "$SRC_DIR/$TARBALL" -C "$BUILD_DIR"

# ---------------------------------------------------------------------------
# 2. Toolchain: Python >= 3.10 (emsdk requirement) and the pinned Emscripten SDK,
#    both installed inside engine/ (gitignored), never system-wide
# ---------------------------------------------------------------------------
python_ok() { "$1" -c 'import sys; sys.exit(0 if sys.version_info >= (3, 10) else 1)' 2>/dev/null; }

PY=""
if command -v python3 >/dev/null && python_ok python3; then
  PY="$(command -v python3)"
elif [[ -x "$HERE/.python/python/bin/python3" ]]; then
  PY="$HERE/.python/python/bin/python3"
else
  case "$(uname -s)-$(uname -m)" in
    Darwin-arm64)  TRIPLE="aarch64-apple-darwin" ;;
    Darwin-x86_64) TRIPLE="x86_64-apple-darwin" ;;
    Linux-x86_64)  TRIPLE="x86_64-unknown-linux-gnu" ;;
    Linux-aarch64) TRIPLE="aarch64-unknown-linux-gnu" ;;
    *) die "no Python >= 3.10 and no standalone build for $(uname -s)-$(uname -m)" ;;
  esac
  log "fetching standalone CPython ${PYTHON_VERSION} (${TRIPLE})"
  mkdir -p "$HERE/.python"
  curl -fsSL "https://github.com/astral-sh/python-build-standalone/releases/download/${PYTHON_RELEASE}/cpython-${PYTHON_VERSION}%2B${PYTHON_RELEASE}-${TRIPLE}-install_only.tar.gz" \
    | tar xz -C "$HERE/.python"
  PY="$HERE/.python/python/bin/python3"
fi
export EMSDK_PYTHON="$PY"

if [[ ! -d "$HERE/.emsdk" ]]; then
  log "cloning emsdk ${EMSDK_VERSION}"
  git clone --quiet --depth 1 --branch "$EMSDK_VERSION" https://github.com/emscripten-core/emsdk.git "$HERE/.emsdk"
fi
if [[ ! -x "$HERE/.emsdk/upstream/emscripten/emcc" ]]; then
  log "installing Emscripten ${EMSDK_VERSION}"
  (cd "$HERE/.emsdk" && ./emsdk install "$EMSDK_VERSION" && ./emsdk activate "$EMSDK_VERSION")
fi
# shellcheck disable=SC1091
source "$HERE/.emsdk/emsdk_env.sh" >/dev/null 2>&1
EMCC_VERSION="$(emcc --version | head -1)"
log "$EMCC_VERSION"

# ---------------------------------------------------------------------------
# 3. Compile the library (UTF-32 widechar so every code point is one unit)
# ---------------------------------------------------------------------------
cd "$LIB_DIR"
log "configure (--host=wasm32-unknown-emscripten --enable-ucs4 --disable-shared)"
emconfigure ./configure --host=wasm32-unknown-emscripten --build="$(./build-aux/config.guess)" --enable-ucs4 --disable-shared --without-yaml --quiet >/dev/null
log "make liblouis.la"
emmake make -C gnulib --quiet >/dev/null
emmake make -C liblouis liblouis.la --quiet >/dev/null

# ---------------------------------------------------------------------------
# 4. Link to an ES module + separate .wasm
# ---------------------------------------------------------------------------
rm -rf "$OUT_DIR"
mkdir -p "$OUT_DIR/tables"
FLAGS=()
while read -r f; do [[ -n "$f" ]] && FLAGS+=("$f"); done < "$HERE/emcc-flags.txt"
log "link: ${FLAGS[*]}"
emcc liblouis/.libs/liblouis.a "${FLAGS[@]}" -o "$OUT_DIR/liblouis.mjs"

WASM_SHA="$(sha256 "$OUT_DIR/liblouis.wasm")"
WASM_NAME="liblouis.${WASM_SHA:0:12}.wasm"
mv "$OUT_DIR/liblouis.wasm" "$OUT_DIR/$WASM_NAME"
MJS_SHA="$(sha256 "$OUT_DIR/liblouis.mjs")"

# ---------------------------------------------------------------------------
# 5. Tables: walk the include chain from tables.list; copy unmodified
# ---------------------------------------------------------------------------
# bash 3.2 compatible (macOS): SEEN is a newline-separated list
SEEN=""
QUEUE=()
while read -r t; do [[ -n "$t" && "$t" != \#* ]] && QUEUE+=("$t"); done < "$HERE/tables.list"
while ((${#QUEUE[@]})); do
  t="${QUEUE[0]}"; QUEUE=("${QUEUE[@]:1}")
  if printf '%s\n' "$SEEN" | grep -qxF "$t"; then continue; fi
  [[ -f "tables/$t" ]] || die "table include chain broken: tables/$t is missing"
  SEEN="$SEEN$t"$'\n'
  cp "tables/$t" "$OUT_DIR/tables/$t"
  while read -r inc; do [[ -n "$inc" ]] && QUEUE+=("$inc"); done < <(awk '$1 == "include" { print $2 }' "tables/$t")
done
TABLES="$(printf '%s' "$SEEN" | sort)"
log "tables: $(echo $TABLES)"

# ---------------------------------------------------------------------------
# 6. Manifest (read by EngineStatus, the worker and validate-data)
# ---------------------------------------------------------------------------
gz() { gzip -9 -c "$1" | wc -c | tr -d ' '; }
bytes() { wc -c < "$1" | tr -d ' '; }
{
  printf '{\n'
  printf '  "liblouis": "%s",\n' "$LIBLOUIS_VERSION"
  printf '  "tarball": { "name": "%s", "sha256": "%s", "url": "%s" },\n' "$TARBALL" "$TARBALL_SHA" "$TARBALL_URL"
  printf '  "emscripten": "%s",\n' "$EMSDK_VERSION"
  printf '  "widechar": 4,\n'
  printf '  "module": { "name": "liblouis.mjs", "sha256": "%s", "bytes": %s, "gzip": %s },\n' "$MJS_SHA" "$(bytes "$OUT_DIR/liblouis.mjs")" "$(gz "$OUT_DIR/liblouis.mjs")"
  printf '  "wasm": { "name": "%s", "sha256": "%s", "bytes": %s, "gzip": %s },\n' "$WASM_NAME" "$WASM_SHA" "$(bytes "$OUT_DIR/$WASM_NAME")" "$(gz "$OUT_DIR/$WASM_NAME")"
  printf '  "tables": [\n'
  first=1
  for t in $TABLES; do
    [[ $first -eq 1 ]] || printf ',\n'
    first=0
    printf '    { "name": "%s", "sha256": "%s", "bytes": %s, "gzip": %s }' "$t" "$(sha256 "$OUT_DIR/tables/$t")" "$(bytes "$OUT_DIR/tables/$t")" "$(gz "$OUT_DIR/tables/$t")"
  done
  printf '\n  ]\n}\n'
} > "$OUT_DIR/manifest.json"

cp "$HERE/LICENSE-liblouis" "$OUT_DIR/LICENSE"
log "done → $OUT_DIR"
cat "$OUT_DIR/manifest.json"